wintercn · May 12, 2021 11:30
diff --git a/parseTree.html b/parseTree.html
 <body>
 <style>
 .number {
    color:purple;
 }
 .keyword {
    color:blue;
 }
 .string {
    color:red;
 }
 #code {
    width:100%;
    height:500px;
 }
 </style>
 <textarea id="code">
 - 章节一
  - 标题一
  - 标题二
    - 子标题三
 - 章节二
  - 标题一
  - 标题二
 </textarea>
 <script>
 var XRegExp =function(xregexps,rootname,flag){
    var expnames = [rootname];
    function buildRegExp(source) {
        var regexp = new RegExp;
        regexp.compile(source.replace(/<([^>]+)>/g,function(all,expname) {
            if(!xregexps[expname])return "";
            expnames.push(expname);
            if(xregexps[expname] instanceof RegExp) return "(" + xregexps[expname].source +")";
            return "(" + buildRegExp(xregexps[expname]).source +")";
        }),flag)
        return regexp;
    }
    var regexp = buildRegExp(xregexps[rootname]);
    this.exec = function(string) {
        var matches = regexp.exec(string);
        if(matches==null) return null;
        var result = new String(matches[0]);
        for(var i = 0; i < expnames.length; i++)
            if(matches[i])
                result[expnames[i]] = matches[i];
        return result;
    }
    this.lastIndex = function (lastIndex) {
        if(arguments.length == 1)
            regexp.lastIndex = lastIndex;
        return regexp.lastIndex;
    }
 }
 function Token(name,value) 
 {
    this.name = name;
    this.value = value;
 }
 function Lexer(source) {
    var tabstop = 2;
    var lex = {
        inputElement:"<whitespace>|<name>|<newline>",
        name:/\- [^\n \t]+/,
        newline:/\n/,
        whitespace:/[ \t]+/,
        comments:/#[^\n]*/
    };

    var input = new XRegExp(lex,"inputElement","g");
    var isLineStart = true;
    var isBlankLine = true;
    var leadingSpace = 0;
    var oldLeadingSpace = 0;
    var dentstack = [];
    this.getToken = function (){
        if(dentstack.length) return dentstack.pop();
        if(input.lastIndex()==source.length) return new Token("ENDMARKER");
        var ele = input.exec(source);
        //console.log(ele);
        if(ele.comments) {
            return this.getToken();
        }
        if(ele.newline) {
            isLineStart = true;
            leadingSpace = 0;
            if(isLineStart) {
                isBlankLine = true;
                return this.getToken();
            }
            return new Token("NEWLINE");
            
        }
        if(isLineStart) {
            leadingSpace = 0;
            if(ele.whitespace){
                for(var i = 0; i < ele.whitespace.length; i++) {
                    if(ele.whitespace.charAt(i) == " ")
                        leadingSpace ++;
                    if(ele.whitespace.charAt(i) == "\t")
                        leadingSpace = Math.floor(leadingSpace/tabstop+1)*tabstop;
                }
            } else {
                dentstack.push(new Token("NAME", ele.name));
            }
            
        
            var dc = (leadingSpace-oldLeadingSpace)/tabstop;
            if(dc<0) 
                while(dc++<0)dentstack.push(new Token("DEDENT"));
            else if(dc>0) 
                while(dc-->0)dentstack.push(new Token("INDENT"));

            oldLeadingSpace = leadingSpace;
            isLineStart = false;
            isBlankLine = false;
            return this.getToken();
        }

        isLineStart = false;
        isBlankLine = false;
        if(ele.name) return new Token("NAME",ele.name);
        if(ele.punctuator) return new Token("PUNCTUATOR",ele.punctuator);
    }
 /*
 NUMBER
 STRING
 NAME
 NEWLINE
 ENDMARKER
 INDENT
 DEDENT
 */
 }

 var lexer = new Lexer(code.value);
 var token;
 </script>
 <pre>
 <script>
 var leadingspace = 0;
 var isLineStart = true;

 var current = {name:"root", children:[]};
 var stack = [current];
 while((token=lexer.getToken()).name!="ENDMARKER")
 {
    console.log(token)
    if(token.name == "NEWLINE") {
        //document.write("<span class='newline'>"+"\n"+"</span>");
        isLineStart = true;
        continue;
    }
    if(token.name == "NAME") {
        //leadingspace+=4;
        //document.writeln("<span class='indent'>"+"indent"+"</span>");

        console.log(stack)
        current = {
            name: token.value,
            children: []
        };
        stack[stack.length - 1].children.push(current);
        continue;
    }
    if(token.name == "INDENT") {
        //leadingspace+=4;
        //document.writeln("<span class='indent'>"+"indent"+"</span>");
        //if(stack[stack.length - 1] !== current)
        stack.push(current);
        //current.children
        continue;
    }
    if(token.name == "DEDENT") {
        stack.pop();
        //leadingspace-=4;
        //document.writeln("<span class='dedent'>"+"dedent"+"</span>");
        continue;
    }
    if(isLineStart) {
        isLineStart = false;
        //document.write("<span class='name'>"+new Array(leadingspace+1).join(" ")+"</span>");
        
    }
 }
 document.write(JSON.stringify(stack[0].children, null, "    "))
 </script>
 </pre>
	<body>
	<style>
	.number {
	color:purple;
	}
	.keyword {
	color:blue;
	}
	.string {
	color:red;
	}
	#code {
	width:100%;
	height:500px;
	}
	</style>
	<textarea id="code">
	- 章节一
	- 标题一
	- 标题二
	- 子标题三
	- 章节二
	- 标题一
	- 标题二
	</textarea>
	<script>
	var XRegExp =function(xregexps,rootname,flag){
	var expnames = [rootname];
	function buildRegExp(source) {
	var regexp = new RegExp;
	regexp.compile(source.replace(/<([^>]+)>/g,function(all,expname) {
	if(!xregexps[expname])return "";
	expnames.push(expname);
	if(xregexps[expname] instanceof RegExp) return "(" + xregexps[expname].source +")";
	return "(" + buildRegExp(xregexps[expname]).source +")";
	}),flag)
	return regexp;
	}
	var regexp = buildRegExp(xregexps[rootname]);
	this.exec = function(string) {
	var matches = regexp.exec(string);
	if(matches==null) return null;
	var result = new String(matches[0]);
	for(var i = 0; i < expnames.length; i++)
	if(matches[i])
	result[expnames[i]] = matches[i];
	return result;
	}
	this.lastIndex = function (lastIndex) {
	if(arguments.length == 1)
	regexp.lastIndex = lastIndex;
	return regexp.lastIndex;
	}
	}
	function Token(name,value)
	{
	this.name = name;
	this.value = value;
	}
	function Lexer(source) {
	var tabstop = 2;
	var lex = {
	inputElement:"<whitespace>\|<name>\|<newline>",
	name:/\- [^\n \t]+/,
	newline:/\n/,
	whitespace:/[ \t]+/,
	comments:/#[^\n]*/
	};

	var input = new XRegExp(lex,"inputElement","g");
	var isLineStart = true;
	var isBlankLine = true;
	var leadingSpace = 0;
	var oldLeadingSpace = 0;
	var dentstack = [];
	this.getToken = function (){
	if(dentstack.length) return dentstack.pop();
	if(input.lastIndex()==source.length) return new Token("ENDMARKER");
	var ele = input.exec(source);
	//console.log(ele);
	if(ele.comments) {
	return this.getToken();
	}
	if(ele.newline) {
	isLineStart = true;
	leadingSpace = 0;
	if(isLineStart) {
	isBlankLine = true;
	return this.getToken();
	}
	return new Token("NEWLINE");

	}
	if(isLineStart) {
	leadingSpace = 0;
	if(ele.whitespace){
	for(var i = 0; i < ele.whitespace.length; i++) {
	if(ele.whitespace.charAt(i) == " ")
	leadingSpace ++;
	if(ele.whitespace.charAt(i) == "\t")
	leadingSpace = Math.floor(leadingSpace/tabstop+1)*tabstop;
	}
	} else {
	dentstack.push(new Token("NAME", ele.name));
	}


	var dc = (leadingSpace-oldLeadingSpace)/tabstop;
	if(dc<0)
	while(dc++<0)dentstack.push(new Token("DEDENT"));
	else if(dc>0)
	while(dc-->0)dentstack.push(new Token("INDENT"));

	oldLeadingSpace = leadingSpace;
	isLineStart = false;
	isBlankLine = false;
	return this.getToken();
	}

	isLineStart = false;
	isBlankLine = false;
	if(ele.name) return new Token("NAME",ele.name);
	if(ele.punctuator) return new Token("PUNCTUATOR",ele.punctuator);
	}
	/*
	NUMBER
	STRING
	NAME
	NEWLINE
	ENDMARKER
	INDENT
	DEDENT
	*/
	}

	var lexer = new Lexer(code.value);
	var token;
	</script>
	<pre>
	<script>
	var leadingspace = 0;
	var isLineStart = true;

	var current = {name:"root", children:[]};
	var stack = [current];
	while((token=lexer.getToken()).name!="ENDMARKER")
	{
	console.log(token)
	if(token.name == "NEWLINE") {
	//document.write("<span class='newline'>"+"\n"+"</span>");
	isLineStart = true;
	continue;
	}
	if(token.name == "NAME") {
	//leadingspace+=4;
	//document.writeln("<span class='indent'>"+"indent"+"</span>");

	console.log(stack)
	current = {
	name: token.value,
	children: []
	};
	stack[stack.length - 1].children.push(current);
	continue;
	}
	if(token.name == "INDENT") {
	//leadingspace+=4;
	//document.writeln("<span class='indent'>"+"indent"+"</span>");
	//if(stack[stack.length - 1] !== current)
	stack.push(current);
	//current.children
	continue;
	}
	if(token.name == "DEDENT") {
	stack.pop();
	//leadingspace-=4;
	//document.writeln("<span class='dedent'>"+"dedent"+"</span>");
	continue;
	}
	if(isLineStart) {
	isLineStart = false;
	//document.write("<span class='name'>"+new Array(leadingspace+1).join(" ")+"</span>");

	}
	}
	document.write(JSON.stringify(stack[0].children, null, " "))
	</script>
	</pre>